Study On Fault Diagnosis And Remaining Useful Life Prediction Methods For Marine Rolling Bearings

In recent years,with the release of "Made in China 2025," ships are gradually being transformed into green,intelligent,and high-end vessels.The "Smart Ship Regulations" issued by the China Classification Society(CCS)categorize smart ships into six major functional modules,including intelligent engine rooms.The regulations specify the monitoring,analysis,and evaluation of the operating status of the ship’s propulsion shaft system in the intelligent engine room.Rolling bearings,as important components of the ship’s shaft system,are also the most prone to failure rotating mechanical parts.The health condition of rolling bearings directly affects the operating status of the shaft system.Once a failure occurs,it may cause irreversible damage to the ship’s shaft system,resulting in significant economic losses and safety issues.Therefore,conducting research on fault diagnosis and remaining useful life prediction of marine rolling bearings has important research value and engineering application significance,The main research content of this thesis can be divided into three parts:(1)A fault diagnosis method based on order analysis and Autogram is proposed to address the strong noise and non-stationary characteristics of vibration signals from rolling bearings in the ship’s shaft system.Firstly,the order analysis method is used to transform the time-domain non-stationary vibration signals into angle-domain stationary vibration signals.Then,the Autogram method is applied to adaptively filter the angle-domain stationary vibration signals,obtaining the order square envelope spectrum,which is compared with the theoretical fault order to achieve fault diagnosis of non-stationary vibration signals.Finally,the proposed fault diagnosis method is validated using the fault signals of variable-speed rolling bearings.Experimental results show that the proposed fault diagnosis method can accurately extract fault feature orders for both inner race faults and outer race faults,effectively achieving fault diagnosis and fault type identification of non-stationary vibration signals from rolling bearings.(2)In order to fully understand the health status of marine rolling bearings and improve the ability of fault prediction,a rolling bearing remaining useful life prediction model based on Convolutional Neural Network combined with Bidirectional Long Short-Term Memory Neural Networkis proposed.Firstly,the rolling bearing vibration signals are envelope demodulated using the Hilbert transform as the input of the prediction model.Then,CNN is utilized to extract the degradation features of rolling bearings,and BiLSTM is employed to learn the relationship between the degradation features and the remaining useful life,thus achieving the prediction of rolling bearing remaining useful life.Finally,the effectiveness of the proposed rolling bearing remaining useful life prediction model is validated using the full life cycle data of rolling bearings.Experimental results demonstrate that the proposed method has good prediction accuracy and can well reflect the performance degradation trend of rolling bearings.(3)According to the research content and practical needs,a fault diagnosis and remaining useful life prediction system for marine rolling bearings is designed and implemented.The system can mainly realize functions such as time-domain analysis,frequency-domain analysis,order analysis,and remaining useful life prediction of rolling bearing vibration signals.The software functions are tested using real ship test data and publicly available datasets.Experimental results show that the software system can effectively analyze,diagnose,and predict the vibration signals of marine rolling bearings.